Download Aging Performance in Crystals - Connor

October, 2008
Author: Jeff Cartright
Crystals Product Manager
The Connor-Winfield Corporation
Aging Performance in Crystals
Introduction
Mass transfer
Aging is the change in crystal frequency over time. This frequency change over the life of the part has to be accounted for in the
application. Understanding the causes of aging, and the methods
used to measure it, enables a method of predicting the rate over
the life (10 to 20 years) of the device.
Quartz crystals change frequency as mass is added or removed
from the wafer. Unaccounted for material on the wafer or in
the package can deposit or leave the surface of the quartz and
change the frequency of the device. In practice, the crystal fabrication process is designed to keep all materials as clean as
possible. If contamination equal in mass to 1-½ monolayers of
Cause of aging
There are two main causes for aging, mass transfer and stress.
quartz is added or removed from the surface, then the frequency
will change 1ppm per megahertz of the device’s frequency [2].
It is critical that the crystal process is as clean as possible to
achieve good aging performance.
Stress
Quartz crystals will change frequency when stress is applied.
This stress can come from any or all of several sources. Package
and mounts, epoxy connection, metal films, and the quartz itself
can be sources of stress. If the stress is constant over the life of
the part, frequency will stay constant. However, in the real world,
stresses relax with time hence, stress related aging occurs.
Package and Mounts
Table 1
In table 1:
A(t) shows stress relief
B(t) shows contamination effects
A(t) + B(t) is the combination.
If a design is dominated by stress, the aging curve will approach
A(t). Likewise, if the design is dominated by contamination, the
aging will approach B(t).
In the real world both mechanisms are present. By looking at the
above curves, it is important to point out the logarithmic nature
of the curves. Crystals age at a fast rate early in their lives, but
tend to settle down with time.
Package materials (metal or ceramic) have different coefficients of expansion than quartz. When quartz is cemented
to the mounts, it is usually cured at an elevated temperature.
When the part is cooled, thermal expansion puts the quartz
under stress. Proper mount design can help reduce this.
Epoxy
Conductive epoxy or glue is used to make electrical connection to the package. During cure, the epoxy shrinks and
puts stress on the quartz through the joint. Softer epoxies
and silicon can be used to reduce this effect. However, the
epoxy (due to outgassing) is a significant source of potential
contamination.
Quartz
Passive vs. Active
Stress in quartz can come from the growing process of the
bars or the machining process of cutting, lapping, and polishing into wafers.
Now that we have established the time and temperature, the
next question is how to collect the data. Two methods are
currently used.
Synthetic quartz is grown in an autoclave from seed material. The quality of the seeds, and the consistency of
temperature and pressure during the growing cycle impact
the number of defects in the bar. These defects, whether
contaminates or dislocations in the lattice structure, can be
points of stress.
The first method is passive aging. Parts are serialized and
measured at room temp in the measurement system before
and after the bake. This is the lowest cost option but it sacrifices accuracy. Passive age can screen fliers and act as a
process control, which alerts the manufacturer to problems.
The second method is active aging. This is performed by
placing each crystal unit in an oscillator circuit so that frequency measurements can be made while the part is in the
bake oven. Measurements can be made several times a day
so that a data curve can be made for each unit. This curve
can be mathematically fit and then projected out over time
allowing for a long-term prediction [7]. Active aging methods result in better accuracy but at a higher cost.
Quartz is a brittle material. The cutting, grinding, and lapping
processes results in small surface cracks coming from the
abrasive particles used in the process. These small cracks
produce stress in the quartz. Processes using chemical
etching can greatly reduce or eliminate this stress in the
surface [3].
Measuring Aging
To improve aging performance, or to insure spec compliance, accurate measurements are key. There are several methods that can
be used, each allowing for trade offs of time, cost, and accuracy.
The true aging of a device can only be measured by placing it in
circuit and frequency over the life of the product. This is unpractical. So a method to simulate 10 or 20 years of life is necessary
to insure aging performance. Accelerated aging at an elevated
temperature was developed to accomplish this.
Accelerated Aging
By increasing the temperature above the normal operating
temperature, aging in crystals can be accelerated. Military
specifications [4] have used 85°C for 30 days and 105°C for
168 hours as an equivalent of one year at 25°C. Other industry standards use 85°C for 1000 hours for first year aging.
The accuracy of the test is better for the longer time, lower
temperature scenario [6]. Connor-Winfield uses 85°C for
1000 hrs as its standard for qualification.
Another benefit of active aging is the effects of driving the
crystal. Actively vibrating the crystal during the test is truer
to real world conditions. Passive aging does not drive the
crystal during the test.
[1] John Vig, “Quartz Crystal Resonators and Oscillators For Frequency Control and
Timing Applications, A Tutorial”, July 2001
[2] John Vig, “Resonator Aging”, Ultrasonic Symposium Proceedings, pg. 848, 1977
[3] R N Castellano, TR Meeker, RC Sundahl, “The Relation Between Quartz Surface
Morphology and the Q of High Frequency Resonators” Frequency Control Symposium, 1977
[4] MIL-PRF-3098
[5] JR Gerhke, R Klawitter, “Experimental Results on Aging of AT-Cut Strip Resonators”,
42 Frequency Control Symposium 1988, pg. 417
nd
[6] M R Miljkovic, G L Trifunovic, V J Brajovic, “Aging Prediction of Quartz Crystal Units”
42 Frequency Control Symposium 1988
nd
Contact
Jeff Cartwright
[email protected]
The Connor-Winfield Corporation
2111 Comprehensive Drive
Aurora, Illinois 60505
Phone: 630- 851- 4722
Fax: 630- 499-2112
www.conwin.com